Mohammad Mahdi Shiraz Bhurwani1,2, Kenneth V Snyder2,3, Muhammad Waqas2,3, Maxim Mokin4, Ryan A Rava1,2, Alexander R Podgorsak1,2, Felix Chin2,3, Jason M Davies2,3, Elad I Levy2,3, Adnan H Siddiqui2,3, Ciprian N Ionita5,6,7. 1. Department of Biomedical Engineering, University at Buffalo, Buffalo, NY, 14228, USA. 2. Canon Stroke and Vascular Research Center, Buffalo, NY, 14203, USA. 3. Department of Neurosurgery, University at Buffalo, Buffalo, NY, 14203, USA. 4. Department of Neurosurgery and Brain Repair, University of South Florida, Tampa, FL, 33606, USA. 5. Department of Biomedical Engineering, University at Buffalo, Buffalo, NY, 14228, USA. cnionita@buffalo.edu. 6. Canon Stroke and Vascular Research Center, Buffalo, NY, 14203, USA. cnionita@buffalo.edu. 7. Department of Neurosurgery, University at Buffalo, Buffalo, NY, 14203, USA. cnionita@buffalo.edu.
Abstract
PURPOSE: Intra-procedural assessment of reperfusion during mechanical thrombectomy (MT) for emergent large vessel occlusion (LVO) stroke is traditionally based on subjective evaluation of digital subtraction angiography (DSA). However, semi-quantitative diagnostic tools which encode hemodynamic properties in DSAs, such as angiographic parametric imaging (API), exist and may be used for evaluation of reperfusion during MT. The objective of this study was to use data-driven approaches, such as convolutional neural networks (CNNs) with API maps, to automatically assess reperfusion in the neuro-vasculature during MT procedures based on the modified thrombolysis in cerebral infarction (mTICI) scale. METHODS: DSAs from patients undergoing MTs of anterior circulation LVOs were collected, temporally cropped to isolate late arterial and capillary phases, and quantified using API peak height (PH) maps. PH maps were normalized to reduce injection variability. A CNN was developed, trained, and tested to classify PH maps into 2 outcomes (mTICI 0,1,2a/mTICI 2b,2c,3) or 3 outcomes (mTICI 0,1,2a/mTICI 2b/mTICI 2c,3), respectively. Ensembled networks were used to combine information from multiple views (anteroposterior and lateral). RESULTS: The study included 383 DSAs. For the 2-outcome classification, average accuracy was 81.0% (95% CI, 79.0-82.9%), and the area under the receiver operating characteristic curve (AUROC) was 0.86 (0.84-0.88). For the 3-outcome classification, average accuracy was 64.0% (62.0-66.0), and AUROC values were 0.85 (0.83-0.87), 0.74 (0.71-0.77), and 0.78 (0.76-0.81) for the mTICI 0,1,2a, mTICI 2b, and mTICI 2c,3 classes, respectively. CONCLUSION: This study demonstrated the feasibility of using hemodynamic information in API maps with data-driven models to autonomously assess intra-procedural reperfusion during MT.
PURPOSE: Intra-procedural assessment of reperfusion during mechanical thrombectomy (MT) for emergent large vessel occlusion (LVO) stroke is traditionally based on subjective evaluation of digital subtraction angiography (DSA). However, semi-quantitative diagnostic tools which encode hemodynamic properties in DSAs, such as angiographic parametric imaging (API), exist and may be used for evaluation of reperfusion during MT. The objective of this study was to use data-driven approaches, such as convolutional neural networks (CNNs) with API maps, to automatically assess reperfusion in the neuro-vasculature during MT procedures based on the modified thrombolysis in cerebral infarction (mTICI) scale. METHODS: DSAs from patients undergoing MTs of anterior circulation LVOs were collected, temporally cropped to isolate late arterial and capillary phases, and quantified using API peak height (PH) maps. PH maps were normalized to reduce injection variability. A CNN was developed, trained, and tested to classify PH maps into 2 outcomes (mTICI 0,1,2a/mTICI 2b,2c,3) or 3 outcomes (mTICI 0,1,2a/mTICI 2b/mTICI 2c,3), respectively. Ensembled networks were used to combine information from multiple views (anteroposterior and lateral). RESULTS: The study included 383 DSAs. For the 2-outcome classification, average accuracy was 81.0% (95% CI, 79.0-82.9%), and the area under the receiver operating characteristic curve (AUROC) was 0.86 (0.84-0.88). For the 3-outcome classification, average accuracy was 64.0% (62.0-66.0), and AUROC values were 0.85 (0.83-0.87), 0.74 (0.71-0.77), and 0.78 (0.76-0.81) for the mTICI 0,1,2a, mTICI 2b, and mTICI 2c,3 classes, respectively. CONCLUSION: This study demonstrated the feasibility of using hemodynamic information in API maps with data-driven models to autonomously assess intra-procedural reperfusion during MT.
Authors: Wade S Smith; Gene Sung; Sidney Starkman; Jeffrey L Saver; Chelsea S Kidwell; Y Pierre Gobin; Helmi L Lutsep; Gary M Nesbit; Thomas Grobelny; Marilyn M Rymer; Isaac E Silverman; Randall T Higashida; Ronald F Budzik; Michael P Marks Journal: Stroke Date: 2005-06-16 Impact factor: 7.914
Authors: Mohammed A Almekhlafi; Sachin Mishra; Jamsheed A Desai; Vivek Nambiar; Ondrej Volny; Ankur Goel; Muneer Eesa; Andrew M Demchuk; Bijoy K Menon; Mayank Goyal Journal: Interv Neuroradiol Date: 2014-02-10 Impact factor: 1.610
Authors: Ryan A Rava; Kenneth V Snyder; Maxim Mokin; Muhammad Waqas; Xiaoliang Zhang; Alexander R Podgorsak; Ariana B Allman; Jillian Senko; Mohammad Mahdi Shiraz Bhurwani; Yiemeng Hoi; Jason M Davies; Elad I Levy; Adnan H Siddiqui; Ciprian N Ionita Journal: J Neurointerv Surg Date: 2020-05-26 Impact factor: 5.836
Authors: Randall T Higashida; Anthony J Furlan; Heidi Roberts; Thomas Tomsick; Buddy Connors; John Barr; William Dillon; Steven Warach; Joseph Broderick; Barbara Tilley; David Sacks Journal: Stroke Date: 2003-07-17 Impact factor: 7.914
Authors: Gregory W Albers; Michael J Wald; Michael Mlynash; Juergen Endres; Roland Bammer; Matus Straka; Andreas Maier; Holly E Hinson; Kevin N Sheth; W Taylor Kimberly; Bradley J Molyneaux Journal: Stroke Date: 2019-09-10 Impact factor: 7.914
Authors: Mohammad Mahdi Shiraz Bhurwani; Muhammad Waqas; Alexander R Podgorsak; Kyle A Williams; Jason M Davies; Kenneth Snyder; Elad Levy; Adnan Siddiqui; Ciprian N Ionita Journal: J Neurointerv Surg Date: 2019-12-10 Impact factor: 5.836
Authors: Romain Bourcier; Mayank Goyal; David S Liebeskind; Keith W Muir; Hubert Desal; Adnan H Siddiqui; Diederik W J Dippel; Charles B Majoie; Wim H van Zwam; Tudor G Jovin; Elad I Levy; Peter J Mitchell; Olvert A Berkhemer; Stephen M Davis; Imad Derraz; Geoffrey A Donnan; Andrew M Demchuk; Robert J van Oostenbrugge; Michael Kelly; Yvo B Roos; Reza Jahan; Aad van der Lugt; Marieke Sprengers; Stephane Velasco; Geert J Lycklama À Nijeholt; Wagih Ben Hassen; Paul Burns; Scott Brown; Emmanuel Chabert; Timo Krings; Hana Choe; Christian Weimar; Bruce C V Campbell; Gary A Ford; Marc Ribo; Phil White; Geoffrey C Cloud; Luis San Roman; Antoni Davalos; Olivier Naggara; Michael D Hill; Serge Bracard Journal: JAMA Neurol Date: 2019-04-01 Impact factor: 18.302
Authors: Mohammad Mahdi Shiraz Bhurwani; Kenneth V Snyder; Muhammad Waqas; Maxim Mokin; Ryan A Rava; Alexander R Podgorsak; Kelsey N Sommer; Jason M Davies; Elad I Levy; Adnan H Siddiqui; Ciprian N Ionita Journal: Proc SPIE Int Soc Opt Eng Date: 2021-02-15
Authors: Jeff Joseph; Benjamin Weppner; Nandor K Pinter; Mohammad Mahdi Shiraz Bhurwani; Andre Monteiro; Ammad Baig; Jason Davies; Adnan Siddiqui; Ciprian N Ionita Journal: Proc SPIE Int Soc Opt Eng Date: 2022-04-04
Authors: Dennis Swetz; Samantha E Seymour; Ryan A Rava; Mohammad Mahdi Shiraz Bhurwani; Andre Monteiro; Ammad A Baig; Muhammad Waqas; Kenneth V Snyder; Elad I Levy; Jason M Davies; Adnan H Siddiqui; Ciprian N Ionita Journal: Proc SPIE Int Soc Opt Eng Date: 2022-04-04